Predicting Performance for Magnetic Fusion | Machine Learning for Fusion

Nuclear fusion is the process that powers the sun and every other active star in the sky. Harnessing it on Earth could mean a revolution in our energy supply, with the promise of safe, clean energy for all.

In this video we dive into a simple but powerful application of machine learning in fusion: predicting the confinement times of magnetic fusion devices.

In the field of fusion, one way to measure how well-confined a fusion plasma is, is with a quantity known as the energy confinement time. Many design features of fusion devices can impact confinement times, and if this impact is well-known, we can build better devices in the future.

In this video we use several machine learning techniques to learn the relationships between design features and confinement for magnetic fusion devices.

The data used is real historical data from tokamak experiments, such as the Joint European Torus in the United Kingdom. The models like the ones we build in the video have been used extensively in fusion science for decades.

So stick along through the video to discover how you can develop simple machine learning models that have important, tangible impacts on the field of fusion.


————————————————————————
📚 Resources
————————————————————————
The tutorial and additional resources for this video can be found here:

https://www.digilab.co.uk/posts/fusio...


————————————————————————
📌 Timestamps
————————————————————————
00:00 - Intro & recap
00:53 - What is confinement?
01:39 - Features that impact confinement
02:20 - The confinement time database
03:03 - JET: The Joint European Torus
03:22 - Importing the data
03:50 - A high dimensional problem
05:27 - Building a linear regression model
07:08 - Testing the linear regression model
08:45 - Building a gaussian process regression model
09:44 - Testing the gaussian process regression model
10:15 - Uncertainty quantification in our models
10:57 - Conclusions

————————————————————————
Connect with me
————————————————————————
digiLab - https://www.digilab.co.uk/

LinkedIn:

  / cyd-cowley-2846a7125  

  / digilab-solutions-ltd  

————————————————————————
🎵 Music
————————————————————————
Achievement by Alex-Productions | https://onsound.eu/
Music promoted by https://www.free-stock-music.com
Creative Commons / Attribution 3.0 Unported License (CC BY 3.0)
https://creativecommons.org/licenses/...

————————————————————————
🏷️ Tags
————————————————————————
Nuclear Fusion
Fusion Energy
Clean Energy
Machine Learning
Artificial Intelligence

————————————————————————
✨ Hashtags
————————————————————————

[#fusion] (https://www.youtube.com/hashtag/fusion)
[#energy] (https://www.youtube.com/hashtag/energy)
[#physics] (https://www.youtube.com/hashtag/physics)
[#megaprojects] (https://www.youtube.com/hashtag/megap...)
[#climatechange] (https://www.youtube.com/hashtag/clima...)
[#engineering] (https://www.youtube.com/hashtag/engin...)
[#science] (https://www.youtube.com/hashtag/science)
[#datascience] (https://www.youtube.com/hashtag/datas...)
[#machinelearning] (https://www.youtube.com/hashtag/machi...)
[#AI] (https://www.youtube.com/hashtag/AI)
[#Artificialintelligence] (https://www.youtube.com/hashtag/Artif...)